Skip to content

Evidence per format

Each reader’s confidence level is computed from the development corpus by the confidence rubric, not assigned by hand. Files counts development files read; confirmed those that match an independent reader (or whose vendor-stored results, such as Chromeleon’s stored peaks, our decoded values reproduce: they confirm those signals only, see Vendor-stored results in the assurance chapter); sources distinct depositors among them; versions distinct format versions, writers and writer versions confirmed; agreement the share of files with an oracle that match it; inferred the share of normalized fields whose meaning was reverse-engineered rather than specified; held-out the generalization benchmark (pass/fail, measured on files never used for development).

format level basis files confirmed sources versions agreement inferred held-out
abf high prior art 37 37 10 19 100% 14% 4/0
agilent-cary medium reverse engineered 24 20 5 1 100% 100% -
agilent-fpa medium prior art 13 13 2 1 100% 57% -
agilent-masshunter high reverse engineered 16 16 6 7 100% 100% -
agilent-seahorse-asyr low reverse engineered 9 3 1 2 100% 100% -
andi-chrom high open spec 25 25 5 3 100% 100% 1/0
applied-biosystems-eds high reverse engineered 23 23 15 11 100% 100% 2/0
arbin-res medium prior art 7 7 4 6 100% 0% -
atf medium vendor docs 7 7 4 1 100% 40% -
bio-rad-pcrd low open spec 0 0 0 0 - - -
biologic-mpr high reverse engineered 21 21 5 6 100% 100% -
biologic-mpt medium reverse engineered 20 19 3 2 100% 100% -
biorad-scn high reverse engineered 13 13 11 5 100% 100% -
blackrock high vendor docs 15 14 3 7 100% 8% -
bruker-bes3t medium prior art 16 14 11 1 100% 0% -
bruker-brml medium reverse engineered 4 4 4 4 100% 0% -
bruker-esp low prior art 3 2 2 2 100% 0% -
bruker-nmr high prior art 50 50 25 7 100% 31% 4/0
bruker-opus high prior art 16 16 5 9 100% 50% 2/0
bruker-raw high prior art 11 11 11 2 100% 0% -
bruker-tdf high prior art 10 10 5 9 100% 56% -
ced-spike2 high prior art 17 16 8 20 100% 20% -
cellvoyager medium prior art 3 3 3 1 100% 82% 0/1
chemstation high prior art 28 28 9 12 100% 96% 3/0
chromeleon medium reverse engineered 12 9 (7 by stored results) 5 3 100% 100% -
cytiva-biacore-blr medium prior art 9 9 2 1 100% 100% -
cytiva-biacore-bme medium prior art 6 6 3 2 100% 100% -
cytiva-unicorn-res low reverse engineered 2 2 2 1 100% 100% -
cytiva-unicorn-zip medium prior art 11 11 4 2 100% 100% -
czi medium prior art 90 84 25 23 100% 30% 3/1
dcimg medium prior art 15 15 3 2 100% 38% -
dm high prior art 89 89 15 15 100% 38% 2/0
emd high prior art 32 32 9 16 100% 76% 1/0
empower-arw low reverse engineered 6 6 1 0 100% 100% -
fcs high open spec 45 45 12 14 100% 26% 4/0
galactic-spc high prior art 12 12 10 5 100% 10% -
gamry-dta medium prior art 6 5 3 3 100% 0% -
genepix-gpr medium reverse engineered 5 5 2 2 100% 0% -
hdf5 low open spec 0 0 0 0 - - -
heka-patchmaster high vendor docs 6 6 5 5 100% 6% -
imagexpress medium prior art 3 3 3 4 100% 82% 1/0
ims high vendor docs 10 10 5 7 100% 91% 1/0
imzml low open spec 2 2 1 3 100% 0% -
intan high vendor docs 12 12 3 4 100% 10% -
jasco-jws high prior art 25 20 6 2 100% 100% -
jcamp-dx high open spec 22 21 4 5 100% 29% 1/0
jeol-jdf high prior art 14 14 8 5 100% 56% 2/0
lif high prior art 22 22 11 5 100% 52% 3/0
magritek-spinsolve medium prior art 26 26 2 4 100% 63% -
malvern-zetasizer-dts low reverse engineered 6 1 1 2 100% 100% -
microcal-itc medium reverse engineered 10 6 2 5 100% 100% -
mirax medium prior art 7 7 1 6 100% 13% -
mrc high open spec 13 13 7 6 100% 24% 2/0
mzml high open spec 96 44 30 16 100% 0% 3/0
mzmlb medium open spec 4 4 1 5 100% 0% 1/0
mzxml high open spec 6 5 4 2 100% 0% -
nd2 high prior art 28 27 7 14 100% 42% 4/0
netzsch-ngb medium prior art 18 18 2 1 100% 0% -
neuralynx high vendor docs 22 22 3 9 100% 30% -
neware-nda medium prior art 5 5 2 2 100% 50% -
neware-ndax medium prior art 4 4 3 3 100% 50% -
nwb high open spec 11 11 11 8 100% 0% 1/0
oib medium prior art 7 7 5 3 100% 43% 1/0
oif low prior art 2 2 1 3 100% 43% -
oir high prior art 16 16 8 5 100% 40% 1/0
ome-zarr high open spec 7 7 5 2 100% 0% -
open-ephys high vendor docs 17 17 3 6 100% 10% -
openlab-cds medium reverse engineered 12 12 3 6 100% 84% -
opera-harmony high prior art 11 11 9 9 100% 82% 2/0
panalytical-xrdml high open spec 7 7 7 11 100% 0% -
perkinelmer-fsm low prior art 2 2 2 0 100% 38% -
perkinelmer-sp medium prior art 8 8 6 0 100% 57% 1/0
plate medium prior art 60 53 14 16 100% 52% 2/1
plexon medium prior art 6 4 1 3 100% 54% -
rdml high open spec 9 9 5 5 100% 4% 1/0
renishaw-wdf high prior art 15 15 6 3 100% 33% 2/0
rigaku-ras medium reverse engineered 4 4 4 2 100% 0% -
rigaku-rasx low reverse engineered 2 2 2 0 100% 0% -
roche-lightcycler-ixo low reverse engineered 4 0 0 0 - 46% -
rotor-gene-rex low reverse engineered 1 1 1 2 100% 46% 1/0
sartorius-octet-frd medium prior art 6 6 2 3 100% 50% -
sciex-wiff medium reverse engineered 7 7 6 9 100% 100% -
ser medium prior art 21 21 4 3 100% 50% 1/0
shimadzu medium reverse engineered 12 9 4 2 100% 100% 0/1
spikeglx high vendor docs 15 15 3 3 100% 20% -
ta-trios medium reverse engineered 12 9 3 5 100% 100% -
ta-universal-analysis medium reverse engineered 7 6 2 1 100% 100% -
thermo-omnic medium prior art 27 27 9 0 100% 29% 2/0
thermo-raw medium reverse engineered 32 31 20 15 100% 61% 2/2
tiff high open spec 113 112 22 38 100% 10% 5/0
varian-nmr high prior art 12 12 6 4 100% 53% 1/0
vsi high prior art 27 11 9 14 100% 100% -
waters-raw high reverse engineered 18 18 6 5 100% 81% -
winwcp low prior art 2 2 1 3 100% 22% -
witec-project high prior art 12 12 7 3 100% 43% -
zvi high reverse engineered 16 15 12 2 100% 100% 1/0

The validated-variant tables themselves (every format version, writer, codec, layout and acquisition mode confirmed, with file and source counts) are in each reader crate’s src/assurance.rs; openreadout info FILE --json → assurance shows which of them a given file uses.